Machine Learning for Developers

Continuous convolution

Convolution as an operation was first created during the 18 century by d'Alembert during the initial developments of differential calculus. The common definition of the operation is as follows:

If we try to describe the steps needed to apply the operation, and how it combines two functions, we can express the mathematical operation involved in the following detail:

• Flip the signal: This is the (-τ) part of the variable
• Shift it: This is given by the t summing factor for g(τ)
• Multiply it: The product of f and g
• Integrate the resulting curve: This is the less intuitive part, because each instantaneous value is the result ...